• Federal Highway Administration >
• Publications >
• Research Publications >
• 07021 >
• Chapt2.Cfm >
• Durability of Segmental Retaining Wall Blocks: Final Report

Durability of Segmental Retaining Wall Blocks: Final Report

CHAPTER 2: LITERATURE REVIEW

2.1 INTRODUCTION

This chapter provides a brief summary of literature to date related to frost resistance of SRW blocks. For conciseness, only an abridged version of a much more comprehensive review by Chan (2006) is presented herein, and readers are directed towards Chan's review (in his 2006 Ph.D. dissertation) for more detailed coverage of freeze-thaw mechanisms (for both conventional concrete and SRWs), including issues related to salt scaling and associated damage. The next section of this chapter briefly discusses some of the more recent literature related to frost damage and salt scaling for conventional concrete, followed by a more thorough summary of published literature on the frost (and salt) resistance of SRW blocks and other dry-cast cementitious materials. Through these discussions, it should be quite clear that the mechanisms of frost damage and salt scaling in conventional concrete are quite complex and not fully understood, and furthermore, that very little is known about freezing and thawing of SRW blocks. This review of available literature confirms the importance and need for research on the frost resistance of SRW blocks, the results of which are discussed in chapter 4.

2.2 FROST DAMAGE IN CONCRETE

2.2.1 Mechanisms of Frost Damage in Concrete

Powers (1949) was one of the first researchers to focus in detail on the mechanisms of frost action and damage in concrete. Powers proposed that the expansion of water during freezing (9 percent volume expansion on freezing) in critically saturated capillary pores forces unfrozen water away from the freezing sites. This displaced water must travel under pressure through the cement paste, and as a result, destructive stresses can occur depending on the amount of resistance to this flow. The role of air voids is to provide "escape" boundaries where the flowing water can escape and freeze without causing damage. Because this theory was based on the flow of water through a permeable cement paste, Darcy's law was employed to model the process. Powers considered a single air bubble and the part of the paste within the bubble's "sphere of influence," shown as a shell in figure 10. This portion of paste contains capillary pores which expel unfrozen water into the air bubble during freezing. Hence, the shell thickness (L) would represent the maximum distance that water must travel before reaching the air void boundary. From Powers' analysis, it was determined that the maximum shell thickness (L_max) above which hydraulic pressures generated by the flow of water is enough to cause cracking of the paste was given by:

      Equation 1

Equation 1

where r_b = radius of bubble

η = coefficient of viscosity

K = coefficient of permeability of the paste

T = tensile strength of the paste

U = quantity of water that freezes per degree drop in temperature

R = freezing rate

It is seen here that increasing the viscosity of the liquid, increasing the freezing rate, decreasing the permeability of the paste or decreasing the strength of paste are more critical as any of these conditions require smaller L_max values (i.e., shorter path lengths).

Figure 10. Diagram. Powers' rendition of an air bubble and its "sphere of influence" (Powers, 1949).

The above theory was, however, developed for a single bubble with its own sphere of influence. To extend this analysis to real paste systems comprised of an assortment of bubble sizes, Powers developed a hypothetical model consisting of equal sized spherical bubbles and derived an expression to estimate the average value of L for all air voids in the paste. This average value was termed the Spacing Factor (), which represents the approximate half-distance between two adjacent air voids. Two expressions of the Spacing Factor were derived based on the relative proportion of paste to air voids. The equations for hence determined were eventually adopted by ASTM C 457 (2004) as a standard method to characterize the air void system in hardened concrete. The full set of equations can be found either in Powers (1949) or ASTM C 457 (ASTM 2004).

Moreover, Powers used experimental data available at the time for a variety of concretes with the intention of estimating the maximum permissible spacing factors for freeze-thaw durable concretes. These concrete mixes had air concretes of about 3 percent, specific surfaces (α) of about 600 square inches/cubic inches and were subjected to freezing rates of [Tf = 9/5 * Tc + 32] degrees Celsius per hour (°C/hr)] (20 degrees Fahrenheit per hour (°F/hr)). Powers determined that the Spacing Factors among the mixes capable of withstanding the imposed freezing conditions were about 0.254 mm (0.010 inch) or less. This value of the spacing factor was in general agreement with the values of L_max computed for the various cement pastes mentioned earlier.

The main advantage of this theory is that it demonstrates the importance of the spacing of air voids in the paste's resistance to freezing and thawing. This theory is also the only one that establishes mathematical relationships between the paste properties, the freezing rate and the spacing of air voids, and how these relate to internal pressures (Pigeon and Pleau, 1995). As will be seen in the next parts however, subsequent to the development of the Hydraulic Pressure Theory, there has been growing experimental evidence that water actually travels towards freezing sites rather than away from them. As a result, doubts have been cast on the actual mechanisms and assumptions entailed by the 1949 Powers' paper. The concepts and equations developed in it are nevertheless still widely used today to evaluate and compare different air void systems.

In the years following the development of the Hydraulic Pressure Theory, Powers, in conjunction with Helmuth, made several important realizations regarding the possible mechanisms taking place during freezing of pastes (Powers and Helmuth, 1953). These new insights, which advanced the understanding of frost damage, were also supported by experimental evidence. The two main aspects suggested by these authors concerned the following:

1. Not all evaporable water in the pores of the paste is freezable. Powers explained that the quantity of ice formed is generally much smaller than the quantity that is thermodynamically freezable at the given temperature. For instance, he quoted other references where it was shown that at −15 °C (5 °F), approximately one-half of the water in pores large enough to be frozen at this temperature remained unfrozen. He attributed this to various reasons. One suggested cause was that due to surface tension, water in the capillary pores would tend to supercool unless it is seeded by an ice crystal on which it can nucleate. At the same time, water in gel pores cannot freeze above -78 °C (108 °F) because it is adsorbed to the surfaces of the calcium-silicate-hydrate (CSH). In fact, at each given temperature below the normal freezing point of water, there are certain pore sizes below which freezing of its water cannot occur. As a result, in a saturated paste, there would always be a fraction of water that is unfrozen, and most of this is likely contained in the smaller pores (including gel pores).
2. Pore water is not pure and contains dissolved chemicals such as alkalis. Ice formation in this solution is accompanied by an increase in the concentration of dissolved chemicals in the unfrozen water. Consequently, concentration differences exist throughout the paste between regions where ice has formed and regions with no ice.

As a result of the above phenomena, a new hypothesis for frost mechanisms and frost damage of pastes was developed as follows. As temperature drops below the normal freezing point of water, ice starts forming in a number of capillary pores while water in gel pores remains unfrozen. The lower free energy (or chemical potential) of ice in the larger pores compared to liquid water in the smaller pores creates a thermodynamically imbalanced condition. This can only be brought back to equilibrium by the flow of liquid water to the ice formation sites. The growth of ice in these larger pores can create enough pressures to damage the paste. Powers referred to these pressures as crystal pressures (Powers, 1975). In addition, the growing chemical concentration of the unfrozen water in the larger pores compared to that in the smaller pores produces a concentration gradient between these two regions. This can also only be balanced by the osmotic flow of liquid water from gel to capillary pores (osmotic pressures). As a consequence, large internal stresses can be generated in the paste due to osmotic pressures from this flow of water. Powers, however, indicated that there is no fundamental difference between the two cases, and they could be simply refereed to as osmotic pressures (Powers, 1975).

It is also noted from these processes that dilution of the unfrozen solution in the capillary pores is possible due to the continued flow of water into the pores. The lower concentration of this solution hence elevates its melting point, thereby promoting more ice formation. Thus, the above described mechanisms become amplified. Furthermore, if the temperature continues to decrease, smaller and smaller pores become prone to freezing and the above phenomena become more pronounced over the entire paste volume.

Powers described experimental observations which agreed with the above mechanisms. It was noted that non-air-entrained pastes would undergo continued dilation while the temperature remained constant. Air-entrained pastes, however, exhibited shrinkage during freezing. Moreover, in other experiments, it was shown that a slightly desiccated non-air-entrained paste would shrink during cooling shortly after the start of freezing, but it would then expand sharply after some low temperature had been passed. These phenomena were explainable considering the mechanisms just described: the paste tends to shrink as water is drawn from the smaller to the larger pores, but would then tend to expand after the capillary voids are filled with water which freezes and expands (Cordon, 1966). These observations could, however, not have been explained using concepts from the Hydraulic Pressure Theory. It is further noted that these length-change observations of freezing concrete specimens have also become the basis for the standard test method, ASTM C 671 "Critical Dilation Test" (ASTM, 2002), which uses length change measurements to compare the frost durability of different concrete mixtures.

As far as air voids are concerned, their role here is to compete with the capillary pores for the flowing water. These air bubbles have more room to allow ice formation, and if spaced sufficiently close together, the air bubbles provide escape routes for the flowing water and alleviate pressure buildup.

Following the early works of Powers (1949) and Powers and Helmuth (1953), a substantial amount of work has been undertaken to advance the understanding of freezing mechanisms and frost damage in porous materials, especially concrete. While hydraulic pressure and osmotic pressure mechanisms are still brought up in various references to explain frost damage, it is evident that other processes may be significant. The application of thermodynamic principles has been vital to elucidate many of these processes. For instance, the expressions relating freeze-point depression to pore size and the Kelvin equation describing capillary condensation in pores have shed light on freezing behavior in various pore sizes (Penttala, 1998). These concepts have set the groundwork for much of the subsequent research on frost processes. Also, the use of low temperature calorimetry has provided a useful way to track ice formation and ice melting. This technique has allowed observing changes in ice formation with changes in the paste microstructure. For instance, it was shown that increasing water-cement ratios and repeated freeze-thaw cycles raise the temperatures at first-freeze (i.e., ice first forms at a warmer temperature). However, more mature pastes and higher cooling rates lower the temperatures at first-freeze. Ice formation was also detected at temperatures down to about −20°C (68 °F) corresponding to freezing of smaller pores. From such tests, the hysteretic nature of freezing and thawing was also elucidated.

As far as frost mechanisms in the paste microstructure are concerned, there is a range of hypotheses proposed in the reviewed literature to explain the various stages occurring during freezing and thawing of pastes and concretes. Kaufmann (2002) provided a qualitative sequential description of these mechanisms. As temperatures drop below the freezing point of bulk water, ice first forms in the larger pores or on the exterior surfaces of concrete. Most of the water in large pores does not interact with the pore walls and is thus held like bulk water. Rapid cooling rates may also induce large hydraulic pressures. Then, as the temperature is further reduced, it is generally agreed that water in smaller pores migrates to larger pores due to the thermodynamic imbalance existing between the unfrozen water in smaller pores compared to the ice in larger pores. This was the basis of the findings in Powers and Helmuth (1953) in which differing length-change trends were also observed in non-air-entrained and air-entrained pastes. While non-air-entrained pastes dilated on freezing, air-entrained pastes shrank on freezing. These results were also confirmed by Penttala (1998). The driving pressures causing this migration of water were determined to be around 1.22 MPa/°C (34.2 MPa/°F) in gel pores (Setzer, 1999) and 1.3 MPa/°C (34.3 MPa/°F) (Scherer and Valenza, 2005). As seen, a few degrees Celsius of undercooling can cause sufficient driving pressures for the redistribution of water in the pore system.

Scherer and Valenza (2005) then pointed out that as the ice in larger pores grow (by "draining" water in smaller pores), this ice eventually exerts crystallization pressures on the pore walls. This pressure is a function of the shape and curvatures in the pores in question as well as the contact angle between ice and pore wall. Scherer attributed this crystallization pressure as being responsible for damage in the material. The amount of pressure generated in the pores has been determined theoretically by Scherer and Valenza (2005) and experimentally by Penttala (1998) who calculated pore pressures based on test chamber relative humidity. In both cases, pore pressures can reach several MPas at several degrees Centigrade below normal freezing point (e.g., at around −5 °C (23 °F)). However, as Scherer pointed out, the calculated pressure applies to one pore only and high stress in a single pore is not likely to crack or fail a material because the volume affected by the stress is too small. Gross material damage is not expected until the crystals propagate through the pore space.

The concept of crusting (Scherer, 1993) and entrapped water (Chatterji, 1999a and 1999b) has not been mentioned frequently in the literature. It deserves attention due to the magnitude of stresses that can be developed under these circumstances. This situation can arise when rapid cooling rates prevail, when a wide range of pore sizes are present (Scherer, 1993), or when dissolved substances continually depress the freezing point of the solution (Scherer, 2005). High pressures in trapped capillary water due to rapid cooling rates was also mentioned in Harnik et al. (1980). The pressures generated in these conditions can reach up to 13.5 MPa/°C (56.3 MPa/°F).

The closed-model container by Fagerlund (1995) which is related to the entrapped water concept revealed that very little freezable water is required to cause damage. As a reminder, the maximum tolerable contents of freezable water values were estimated to be 5 percent of the cement volume at −5 °C (23 °F), 2 percent at −10 °C (50 °F) and 0.7 percent at −20 °C (68 °F). As mentioned before, this model is highly conservative due to the assumption that all water is contained inside the sphere. However, in separate calculations, Fagerlund also demonstrated that substantial amounts of freezable water can potentially exist in pastes, even those that are considered dense such as those with low water-cement ratio. As such, there is always a potential for freeze-thaw damage as long as water is available. It should also be noted that pastes subjected to freeze-thaw cycles have the potential to acquire more water with each cycle due to the micropump effect (Setzer, 1999). This water uptake has been observed experimentally by various researchers who concluded that the uptake caused by freezing and thawing can be even higher than the uptake from capillary absorption alone. Thus, the degree of saturation in the paste gradually increases. It would therefore not be surprising to observe durable behavior in cement pastes during early freeze-thaw cycles but then observe increasingly higher vulnerability with increased number of cycles.

2.2.2 The Role of Deicing Salts in Frost Damage

The above discussions have focused on classical and modern theories on frost damage in concrete; however, these discussions have not yet delved into the role and importance of deicing salts on deterioration in cold climates. It is well established that the presence of deicing salts in concrete can greatly affect its freeze-thaw durability. Harnik et al. (1980) pointed out that concretes generally exhibit lower resistance to the combined effects of frost and salts compared to frost alone. The reasons for the poorer performance in the presence of salts are not fully understood. It is nevertheless recognized that the presence of salts in solution has three fundamental effects: it lowers the vapor pressure of the solution, it depresses the freezing point of the solution (Pigeon and Pleau, 1995), and it increases the viscosity of the solution. These effects are likely responsible, at least in part, for the more severe damage to concrete. Moreover, from a phenomenological standpoint, it has been shown that salts are most damaging to concrete surfaces at concentrations of about 3 to 4 percent depending on the particular deicer (Verbeck and Klieger, 1957).

Various theories have been proposed to explain the deleterious effects of salts. These have been presented in Harnik et al. (1980) and in Pigeon and Pleau (1995) and are summarized here.

Amplification of Osmotic Pressures

The presence of deicing salts can give rise to local concentration gradients causing an osmotic imbalance. The resulting osmotic pressures would amplify those already existing due to ice-water thermodynamic imbalance (according to Powers and Helmuth's osmotic pressure theory). Salts on the surface of concrete would also enhance the vapor pressure differential between surface ice and supercooled pore water. Based on Litvan's desorption theory, the pore water would thus have a greater tendency to migrate out of the concrete. In both cases, osmotic pressures are most likely increased.

Degree of Saturation

The presence of salts in solution lowers the equilibrium vapor pressure of the solution. As a result, water molecules in the vicinity of the solution have a greater tendency to migrate towards the solution compared to pure water. This is the basis for the hygroscopicity of saline solutions. The degree of saturation in the concrete is thus increased.

Supercooling

Due to the depression of freezing point, ice crystals do not form on the surface of concrete at temperatures near 0 °C thereby causing supercooling of the pore water. When this water eventually freezes, the phase transition effects are more destructive than in normal freezing. Harnik et al. (1980) experimentally demonstrated that large supercoolings lead to faster propagation of the ice front and thus greater hydraulic pressures.

Thermal Shock

When salts melt ice on the surface of concrete, the endothermic phase transition can draw up large amounts of heat, primarily from the concrete itself. The sudden extraction of heat can cause shock-like cooling with consequent high tensile stresses at the concrete surface. These tensile stresses may be large enough to rupture the outer layers of the concrete.

Layer-by-Layer Effect

The presence of salts lowers the freezing point of the solution. If the salt concentration is non-uniform in concrete, there would also be non-uniform freezing throughout the concrete. These various regions would thus have different dilation properties from which stresses can develop. Variations in dilation could arise from differences in ice formation or from contraction of the paste (if protected by air voids).

Salt Crystallization

Salt crystal growth in large pores in concrete can occur if the salt solution becomes supersaturated. Supersaturation can arise due to evaporation of water from the solution, transport of salt ions from smaller pores toward salt crystals in larger pores, freeze concentration of the salt solution, or solution reaching eutectic conditions. The continued growth of salt crystals in pores can exert sufficient pressures on the pore walls to cause damage.

The above-described mechanisms can account at least in part for the severity of damage caused by salt solutions (i.e., any of the above or combination of the above may be responsible for the damage). It is also noted that due to the depressed freezing point of solutions, salts are beneficial to concrete by delaying ice formation. The countering of positive effects (i.e., delayed ice formation) and negative effects (described above) may be the reason behind the pessimum salt concentrations observed by Verbeck and Klieger (1957) (about 3 to 4 percent). At high concentrations, the positive effects could outweigh the damage potential of the salts, while at low concentrations, the effect of the salts may not be significant. Regardless of the damage mechanisms, it has been demonstrated that air entrainment can significantly improve the deicing scaling resistance of concrete (Pigeon and Pleau, 1995).

2.3 FREEZE-THAW DURABILITY OF DRY-MIXED CONCRETE PRODUCTS

2.3.1 Introduction

SRW units are generally considered a type of dry-mixed concrete product. Other types of dry-mixed concrete products may include concrete masonry units, concrete pavers, and roller compacted concretes. Pigeon and Pleau (1995, p. 206) define dry concretes as "[concretes] in which the amount of water or cement paste in the mix is significantly lower than that in normal concretes." In these mixtures, the amount of water is carefully controlled because the stiffness of the mix plays an important role in the placement process. For instance, SRW units are demolded immediately after they are compacted, and thus a high stiffness is required for the unit to retain its shape after demolding (figure 11).

Figure 11. Photo. Demolding of SRW units during production.

Due to the low paste content of dry concretes, the void spaces between aggregate particles cannot be filled completely. Consequently, a network of irregularly shaped voids is created as was previously shown in figure 7, which compares the internal structures of ordinary and SRW concretes. These voids are termed "compaction voids" since they are formed during the compaction process (Pigeon and Pleau, 1995). The role of these voids in frost durability of dry concretes is still uncertain as will be described in the following sections. It is interesting to note that dry-mix shotcrete which also contains low water-cement ratios and is compacted pneumatically, has a very low volume of compaction voids. The cement content in shotcretes is in the range of 400 to 500 kilograms per cubed meter (kg/m³) (674 to 843 lbs/yd³), as batched (Morgan, 1995), which is higher than the 250 to 380 kg/m³ (421 to 641 lbs/yd³)for frost resistant dry-mixed concretes (see following sections). Pigeon and Pleau (1995) also explain that due to the low water-cement ratio in dry concretes, there is less mixing and dispersion of cement grains. The presence of unhydrated cement particles may influence the frost durability of dry concretes (MacDonald et al., 1999).

The literature on SRW frost durability is fairly limited to date, because SRW construction has only become popular in recent years, and frost-related problems have only surfaced recently (SEM, 2001). There is a larger body of literature for other more established dry concrete products such as the ones mentioned earlier. Hence, this section of the literature survey examines frost-related research work performed on various types of dry-mixed concrete products.

2.3.2 Mechanisms of Freeze-Thaw Damage in Dry-Mixed Concretes

The theories of ice formation mechanisms and frost-induced damage presented in sections 2.2.1 and 2.2.2 were developed primarily for conventional structural concretes. For these concretes, ice was generally presumed to form in capillary voids, and the role of the larger air voids was to provide escape boundaries where ice could freely grow. Whether these mechanisms apply similarly to dry-mixed concretes is unresolved, largely because of the different microstructure exhibited by dry-mixed concretes compared to conventional concretes.

Detailed surveys of existing literature on frost durability of dry-mixed concretes are covered in Haisler (2004), Hance (2005) and SEM (2001). Since the body of literature on this subject is not extensive compared to ordinary concretes, there is a fair amount of overlap in these reviews comprising approximately 25 separate investigations on various types of dry-mixed concrete products. A summary of the findings is presented here.

The reviewed literature covered a wide range of dry-mixed concrete products, focusing primarily on the influence of mix composition and material properties on the frost durability of these materials. Among the various types of concrete products investigated were roller-compacted concretes (RCC), concrete masonry units, concrete paving units and SRW units that were evaluated using a variety of freeze-thaw test methods. For instance, in some studies, RCC specimens were evaluated using ASTM C 666 (2004), Procedure A methods (2004). This method involves rapid freezing and thawing of specimens fully submerged in water. In other investigations, SRW units were evaluated using ASTM C 1262 methods (2003), which involves freezing and thawing of specimens partially immersed in water. Other investigations may have involved testing concrete masonry units using ASTM C 672 methods (2004), which also involves freezing partially immersed specimens. The freezing times are, however, much longer (20 ± 1 hour), compared to that required in ASTM C 1262 (2004). As a result, it is not surprising that a fairly large range of results, observations, and performance criteria have been garnered from these investigations.

Although the reviewed literature does not point to a single frost damage mechanism in dry concretes or to a single frost durability predicting parameter, there is some general agreement with respect to certain factors affecting frost resistance. Higher compressive strengths, lower water-cement ratios, and lower absorptions have generally been observed to decrease freeze-thaw vulnerability. Specific values of these parameters depended largely on the specific material tested and the freeze-thaw method employed. For example, a minimum compressive strength of 21 MPa (3,040 psi) was suggested for concrete masonry units under "severe" exposure, while minimum compressive strengths of 50 MPa (7,250 psi) had been suggested for "durable" concrete paving blocks. For these same paving blocks, a water-cement ratio below approximately 0.30 was required for frost durability. In a separate study on concrete pavers, minimum compressive strengths in the range of 55 to 67 MPa (8,000 to 9,700 psi), with accompanying absorption of less than 4 percent by mass, were recommended for frost durable material. In other studies on SRW units, it was shown that units displaying water absorption lower than 176 kg/m³ had better likelihood of meeting the ASTM C 1372 (2001) freeze-thaw criterion of 1 percent maximum mass loss. In this same study, units with compressive strength higher than 62 MPa (9,000 psi) were also more likely to meet this 1 percent maximum mass loss criterion. Despite observing general trends in the durability of dry concrete products with respect to the above-described material properties, both Haisler (2004) and Hance (2005) concurred that these properties were weakly correlated to frost durability. As such, the adequacy of any of these properties as a reliable predictor of freeze-thaw performance is questionable. From their reviewed literature, these authors did find, however, that cement content was an important parameter for frost durability. Recommended minimum values ranged from 252 to 395 kg/m³.

In the literature reviewed by SEM (2001), other factors were also investigated for their relevance to frost durability. As far as mix composition is concerned, conflicting results have been reported. While binder type and water-binder ratio were reported to be of little influence on the frost resistance of concrete pavers, the use of mineral admixtures (e.g., silica fume) was found to improve frost durability of concrete masonry units. Aggregate selection and proper curing were also mentioned as being important for the frost resistance of dry concretes. In particular, it has been reported that for SRW units, larger quantities of unhydrated cement particles existed in frost susceptible units compared to more durable units. As a result, the degree of hydration (as influenced by curing methods) was reported to be critical for SRW units.

2.3.3 The Role of Salts in Frost Damage in Dry-Mixed Concretes

In their survey of literature on frost durability of dry concretes, SEM (2001) also covered several studies related to the deicing salt scaling resistance of these concretes. This compilation by SEM comprised five separate investigations covering concrete pavers, concrete masonry units and SRW units. One of the key points identified in this survey was the fact that manufacturing processes can strongly affect the scaling resistance of paving blocks. In one study, it was noted that "it was extremely difficult to accurately evaluate the durability of the paving blocks since all productions were plagued by large variations in the properties of the paving blocks." In a similar study, it was noted that special care should be taken to "adequately consolidate" all pavers since their performance could be "strongly affected by the casting operations."

Regarding correlations to other properties, it was noted from one study that the scaling resistance of paving blocks correlated well with capillary water absorption, but less so with compressive and flexural strengths. However, in another study, no useful correlations were found to exist between the durability of the specimens and parameters such as compressive strength, dynamic modulus of elasticity, and water absorption. Other investigations also pointed toward the importance of specifying minimum cement contents that were determined to be in the range of 320 to 380 kg/m³ to ensure durable paving units. For these same units, an average compressive strength of 65 MPa (9,400 psi), an average tensile strength of 6 MPa (870 psi), a mean unit weight of about 2275 kg/m³ and a mean water absorption of 4 percent were typical of durable units. This same study indicated that the cement-aggregate ratio strongly influenced the frost durability, although no values of this parameter were provided. Moreover, a maximum water-cement ratio of 0.35 was demonstrated to be a common quality among durable units.

2.3.4 Role of Air and Compaction Voids

In ordinary concretes, it is well-established that the presence of microscopic, discrete, and well-dispersed air voids helps reduce the damage caused by repeated cycles of freezing and thawing. In the case of dry-mixed concrete products however, the use of mixture designs differing from those of ordinary concretes, and the batching, casting, and curing methods produce a material that exhibits a fairly dissimilar microstructure compared to that in ordinary concretes. As a result of this, two issues arise:

• Difficulty in developing an air void system in dry-mixed concrete products.
• The role of air and compaction voids with respect to frost protection.

This section briefly reviews several recent studies and investigations which address these two issues. It will be shown that conflicting results arise and that there are no simple answers to the above issues.

2.3.4.1 Air Void Characteristics in Low-Slump Concretes (Whiting, 1985)

Whiting investigated the air void characteristics in fresh and hardened low-slump dense concretes (LSDC) which were used as overlays for highway bridge decks. This type of concrete nominally incorporates cement contents of approximately 490 kg/m³, water-cement ratios of 0.30 to 0.32 by weight and air contents of 6.5 ± 1.0 percent. These mixes require vigorous vibration either by vibration or rodding for its consolidation. Although the cement content at about 250 to 400 kg/m³ is higher than that in dry-concrete products, it is difficult to entrain air voids into the stiff LSDC mixture. Hence, Whiting's study focused on two objectives: establishing dosages of various air-entraining admixtures required to achieve specified air contents in the freshly mixed LSDC, and investigating the air void system of LSDC mixes meeting the specified air content.

With respect to entraining air voids, Whiting found that using neutralized vinsol resin and alkyl-benzyl sulfonate-type air entraining agents, large dosages were required to achieve the specified air contents (6 ± ½ percent) in LSDC. The required dosages of these admixtures were up to 10 times higher than that required in ordinary concretes. Once reaching its target value, the air content became less sensitive to changes in the admixture dosage. On the other hand, attempts to use other types of air entraining admixtures (alkali-stabilized, saponified natural wood resin and an organic acid salt consisting of tall-oil derivatives) were not successful in achieving the target air content, regardless of dosage.

Hardened concrete air void parameters were examined for mixes meeting the specified 6 ± ½ percent in the fresh state. For these mixes, void frequencies larger than 10 per inch, specific surfaces exceeding 39 mm²/mm³ (1,000 square inch/cubed inch) and spacing factors, below 125 microns (μm) (0.005 inch) were consistently obtained. The specific type of air entraining admixture used did not appear to affect these results (for mixes achieving 6 ± ½ percent air content). Whiting also reported that these mixes exhibited hardened air contents that were approximately 1 to 2 percent lower than those measured in the fresh state. The size distribution of these air voids was also found to be finer compared to that in ordinary concretes.

2.3.4.2 Frost and Salt Scaling Resistance of RCC (Marchand et al., 1990)

Marchand et al. (1990) conducted an investigation of the frost durability and characteristics of the air void system of RCC. A total of 20 RCC loads were produced in a field test site encompassing the following variables in mix composition:

• Binder type including ASTM Type I, ASTM Type III and silica fume addition.
• Water-cement ratios of 0.27, 0.33 and 0.35.
• Cement contents in the range of 12 to 16 percent.

For the air-entrained mixes, an aqueous solution of neutralized sulfonated hydrocarbon was used as air-entraining admixture. This was added at twice the manufacturer recommended dosage, and a total mixing time of 5 minutes was maintained. For all mixes, ASTM C 666 (Procedure A, ASTM 2004) rapid freeze-thaw tests, ASTM C 672 (2004) salt scaling tests and ASTM C 457 (2004) hardened concrete air void analyses were conducted.

Marchand et al. found that the addition of an air-entraining admixture did not assist in the entrainment of air bubbles, even with higher than normal dosages. Most of the voids found were of the compaction type. It is not clear from this paper, however, how air voids and compaction voids are distinguished from one another. Large variations in total air content were determined, from as low as 2 to 3 percent to as high as 10 percent. Values of the specific surface for most mixes were lower than 25 mm^-1, while spacing factors were generally found to be less than 250 μm (0.010 inch). Marchand questioned the validity of applying ASTM C 457 (2004) parameters to this type of concrete, in light of the irregular shape of compaction voids observed.

As far as frost resistance is concerned, all RCC samples tested withstood 300 cycles of freezing and thawing in water without any significant deterioration. Consequently, Marchand suggested that some compaction voids may act as air voids, but that this positive influence of compaction voids should not be relied upon too heavily. The deicing salt scaling resistance of the RCC mixes was found to be poor. Reasons provided for this include interconnected compaction voids which favor saturation of the concrete, nonhomogeneity of the paste, and lack of air bubbles. From these results, Marchand also suggested that it is possible that the well-established relationship between spacing factor and freeze-thaw durability in conventional concretes does not apply to RCC.

2.3.4.3 Air Entrainment in No-Slump Mixes (Marchand et al., 1998)

In a separate laboratory study, Marchand et al. examined 21 different zero-slump concrete mixtures for their air-entrainment characteristics and hardened air void parameters. In all mixes, the cement content was fixed at 13 percent of the total mass of dry materials and the water-cement ratio maintained at 0.37. Two types of mixers were used: a counter-current pan mixer and a revolving drum mixer. Four types of air-entraining agents were used: synthetic detergent, neutralized sulfonated hydrocarbon salt, vinsol resin, and vegetable oil extract. These were added at dosages of 1 to 4 milliliters per kilogram (ml/kg) (1.5 to 6.1 fl oz/100 lbs of cement, which in some cases represented more than 10 times the manufacturer recommended dosage. These mixes were evaluated by ASTM C 457 Modified Point Count method (ASTM 2004), pressure saturation tests (to measure the volume of capillary voids and the amount of non-connected voids) and scanning electron microscopy.

With respect to ASTM C 457 (2004) air void analysis, distinction was made between spherical air voids and compaction voids. It was determined that the spherical air void content was generally low (less than about 1.5 percent). Mixes exhibiting higher spherical air void content corresponded to those mixed in the pan mixer and employing synthetic detergent or vegetable oil extract at 4 ml/kg of cement. The total air content of the mixes was quite variable, ranging from 5.4 to 10.8 percent (by volume). The authors concluded that entraining air bubbles in no-slump mixes is difficult due to low water content in the mix. Even so, air entrainment is not impossible, depending on the type of mixer and air-entraining agent used. Scanning electron microscopy observations revealed the presence of microscopic bubbles (less than 50 μm (0.002 inch)) in the air-entrained mixes. The role of these microscopic bubbles in providing frost resistance to the concrete remained uncertain. As far as absorption was concerned, the capillary absorption was found to be generally below 5 percent despite the presence of compaction voids. Pressure saturation test results revealed that a certain percentage (3 to 4 percent) of voids were nonconnected. Hence, the authors suggested that part of the nonconnected voids could act as air bubbles during freezing.

2.3.4.4 Air Entrainment in Dry Masonry Concrete (Hazrati and Kerkar, 2000)

Hazrati and Kerkar studied the freeze-thaw durability of concrete masonry units containing an integral water repellent admixture and a "novel freeze-thaw admixture." Several mixes were produced in which the cement-aggregate ratio ranged between 11 to 18 percent, and the dosage of the above-mentioned admixtures was varied. These mixes were evaluated using the following test methods: ASTM C 1262 (2003) freeze-thaw durability in water and in 3 percent sodium chloride (NaCl) solution, ASTM C 140 (2000) compressive strength, water absorption and unit weight and ASTM C 457 (2004) Modified Point Count method.

From the freeze-thaw tests, it was generally observed that mixes containing only the integral water repellent admixture exhibited early deterioration, regardless of the cement content. Samples tested in water surpassed a 1 percent mass loss after about 30 cycles, while samples tested in saline solution showed substantial mass loss (>5 percent) after only 10 or less cycles. However, when the "novel freeze-thaw admixture" was added, the mixes showed much improved durability. The samples tested in water exhibited mass loss less than 1 percent even after 120 cycles, while samples tested in saline solution displayed less than 1 percent mass loss even after 80 cycles. With respect to air void systems, Hazrati and Kerkar reported spacing factors of 450 to 550 μm (0.018 to 0.022 inch) for non-air-entrained mixes and 200 to 300 μm (0.008 to 0.011 inch) for air-entrained mixes. From this the authors suggested maximum spacing factors of 300 μm (0.011 inch) for frost resistance in water and 200 μm (0.008 inch) for frost resistance in saline solution. Image analysis on fluorescent impregnated thin sections revealed that the degree of hydration in mixes containing only a water repellent admixture was below 50 percent. On the other hand, mixes containing the "novel freeze-thaw admixture" displayed degrees of hydration of about 70 percent. Specimen age or curing methods were not provided in this reference.

From their results and observations, the authors concluded that the "novel freeze-thaw admixture" significantly improved frost resistance in water and in saline. The beneficial effect of using this admixture was even greater than that of increasing cement content alone. This benefit was likely a result of entraining air voids into the concrete and dispersing cement grains to allow further hydration. With respect to the lower durability found in mixes with water repellent admixture, Hazrati and Kerkar (2000) suggested that "conditions that reduce permeability without drastically decreasing the porosity could be detrimental to the freeze-thaw durability of cement based materials."

2.3.4.5 Other Studies (Pigeon and Pleau, 1995 and SEM, 2001)

Pigeon and Pleau (1995) provided a brief compilation on the work done by several other researchers on the issue of air entrainment in dry concretes. They cited three separate studies in which it has been possible to introduce air bubbles to such concretes. In one study on RCC, the batching sequence had to be altered by first mixing the cement, water, air entraining agent and a portion of the aggregate. Once mixing had been carried out long enough to allow air voids to form, the rest of the aggregates were added. In the other two studies, high-energy mixers consisting of rotating blades were employed to obtain air entrainment in the mix.

SEM (2001) also covered several other investigations related to frost protection in dry concrete mixtures. With respect to RCC tested according to ASTM C 666 (Procedure A, Rapid Freezing and Thawing in Water, ASTM 2004), one study showed that the addition of air-entraining agents had a positive effect on frost durability, although there was no significant influence from the particular type of agent used. In another study, it was concluded that the frost durability of RCC was directly related to the air void spacing factor, with a maximum suggested value of 250 μm (0.010 inch). This same study also suggested showed that non air-entrained RCC could be to a certain degree resistant to frost. Hence, it was concluded that compaction voids could offer similar frost protection to concrete as entrained air voids.

2.3.4.6 Summary of Studies on Air Entrainment and Compaction Voids

In general, it is apparent that issues relating to air entrainment and frost protection are distinct for dry-mixed concretes compared to ordinary concretes. In ordinary concretes, the use of air entrainment has been shown to improve frost resistance and deicing-salt scaling resistance. The necessity of air voids for frost protection in dry concretes is still contested, however. The issue of air entrainment itself has led to conflicting views. While some studies showed that air entrainment is extremely difficult to achieve regardless of the dosage of admixture used, other studies showed that air entrainment was possible under certain conditions. It appears that the specific type of air-entraining agent used, the mix composition, the type of mixer used and perhaps even the batching and mixing procedures employed played important roles in determining the success in entraining air voids. At the other end of the spectrum, Hazrati and Kerkar (2000) showed that with nominal amounts of a "novel freeze-thaw admixture," good air void systems could be achieved and frost durable masonry units were obtained.

The roles of air and compaction voids in providing frost protection to dry concretes still remain unexplained. Marchand et al. (1998) reported that while some field and laboratory investigations tended to indicate that some compaction voids can act as air voids and offer frost protection, other reports have demonstrated that non-air-entrained, no-slump concretes are vulnerable to frost damage. The connectivity of these compaction voids has been cited as being a critical parameter, since some researchers suggested that isolated compaction voids may act as air voids. On the other hand, connected voids can increase saturation and exacerbate frost damage.

The validity of ASTM C 457 (ASTM 2004)parameters in characterizing the void system of dry concretes has also been questioned by several authors. This is because the equations in ASTM C 457 (ASTM 2004) were developed assuming spherical voids uniformly dispersed in the cement paste. This hypothetical void shape and spatial distribution is even less valid in dry-mixed concrete than in conventional concrete. In dry-mixed concretes, interconnected compaction voids may dominate. Consequently, the relationships developed between spacing factor and freeze-thaw durability for conventional concretes may not be valid for dry concretes. Hence, it has been suggested that the actual role of compaction voids during freeze-thaw conditions must be first understood before establishing any relationships between ASTM C 457 (ASTM 2004) parameters and frost durability in dry concretes.

2.4 SUMMARY

This chapter summarized some of the main published works to date on the frost durability of conventional concrete, SRW blocks, and other dry-mixed concrete products. Through this discussion, it is quite evident that the mechanisms of frost damage and salt scaling are not completely understood for conventional concrete, and this understanding is even less when considering SRW blocks and other dry-cast products. Lack of understanding with regard to SRW block durability can be attributed to several factors, including the relative newness of the SRW market (compared to conventional concrete), the unique nature of SRW block microstructure, and the general lack of scientific publications on the topic. Based on this review, the need for comprehensive research on the frost resistance of SRW blocks is quite evident, and the efforts detailed in the rest of this report aim at addressing these needs.